jpmsilva1
- 16 skills
- 0 followers
- 5 hours ago last updated
- ▌ Eval Harness · jpmsilva1Formal evaluation framework for Claude Code sessions implementing eval-driven development (EDD) principles
- ▌ Mle Workflow · jpmsilva1Production machine-learning engineering workflow for data contracts, reproducible training, model evaluation, deployment, monitoring, and rollback. Use when building, reviewing, or hardening ML systems beyond one-off notebooks.
- ▌ Pytorch Patterns · jpmsilva1PyTorch deep learning patterns and best practices for building robust, efficient, and reproducible training pipelines, model architectures, and data loading.
- ▌ AI Regression Testing · jpmsilva1Regression testing strategies for AI-assisted development. Sandbox-mode API testing without database dependencies, automated bug-check workflows, and patterns to catch AI blind spots where the same model writes and reviews code.
- ▌ Scientific Thinking Literature Review · jpmsilva1Systematic literature-review workflow for academic, biomedical, technical, and scientific topics, including search planning, source screening, synthesis, citation checks, and evidence logging.
- ▌ Scientific Thinking Scholar Evaluation · jpmsilva1Structured scholarly-work evaluation for papers, proposals, literature reviews, methods sections, evidence quality, citation support, and research-writing feedback.
- ▌ Lint Vault · jpmsilva1Triggered when the user types /lint or asks to health-check the Obsidian Vault. Scans the wiki/ directory to identify structural issues, orphan pages, broken wikilinks, stale index entries, and contradictions.
- ▌ Save Session · jpmsilva1Triggered when the user types /save or asks to save the session. Generates a session log and saves it to the Obsidian Vault.
- ▌ Output Shaper · jpmsilva1Dynamic token compression and verbosity controller. Forces the agent to be incredibly concise, acting as a manual output shaper. Supports three intensity profiles: lite, balanced, and ultra. Use whenever the user types "/output-shaper", complains about verbosity, asks to save tokens, or wants shorter answers.
- ▌ Resume Session · jpmsilva1Triggered when the user types /resume or asks to resume. Reads the previous session logs from the Obsidian Vault to restore context.
- ▌ Paper Code Finder · jpmsilva1Find source code implementations and GitHub repositories for academic AI/ML research papers. Use this skill whenever a user asks for the code, repository, huggingface, or implementation of a paper. Ensure you trigger this even if they just say "is there code for this?" while looking at a paper.
- ▌ Experiment Sweeper · jpmsilva1Expert in ML hyperparameter orchestration. Converts scripts to use Hydra/OmegaConf and sets up Weights & Biases Sweeps. Authored by João P. M. Silva.
- ▌ Research Orchestrator · jpmsilva1Triggered when the user types /research or asks to start a research workflow. Guides the user through the full academic research lifecycle by suggesting the right skills at each stage. Opt-in only.
- ▌ Academic Code Replicator · jpmsilva1 bundleFull-lifecycle skill for replicating experiments from academic papers. Guides the agent through understanding the original codebase, constructing isolated environments, making safe platform-compatibility adaptations, executing on available hardware, and producing a documented comparison against paper-reported results. Use this skill whenever the user says things like: "replicate this paper", "reproduce this experiment", "make this old code run", "run this paper's code", "I'm trying to get this GitHub repo from a paper working", "help me replicate these results", "this code is from 2019 and won't install", "I need to reproduce Table X from this paper", or any variation of scientific code replication. PROACTIVELY trigger even if the user just says "help me run this paper's experiments" without explicitly mentioning replication.
- ▌ Distributed Gpu Engineer · jpmsilva1Expert in scaling ML training across multiple GPUs and nodes. Masters SLURM, PyTorch Distributed Data Parallel (DDP), Ray, and CUDA OOM debugging. Authored by João P. M. Silva.
- ▌ Academic Rebuttal Simulator · jpmsilva1Simulates 'Reviewer 2' for ML papers (NeurIPS, ICLR). Critiques methodology, novelty, baselines, and related work. Outputs structured OpenReview-format reviews with sub-scores, suggests target venues with verified acceptance rates, and drafts conference rebuttals. Authored by João P. M. Silva.